ELLIPSIS IN SEQUENCES OF MULTIPLE
WH-WORDS
64 / 2016 / 1 Hana Gruet-Skrabalova
ČLÁNKY – ARTICLES
Abstract
This article deals with elliptical sequences of multiple wh-words in Czech. It is argued that such
sequences involve two different constructions, elliptical multiple questions and elliptical assertions. Wh-words in elliptical multiple questions share the properties of wh-words in non-elliptical
questions; they have the same syntactic distribution and semantic interpretation, they do not
show superiority effects, and their combinability depends on their reading. Moreover, elliptical
questions can be reconstructed into syntactically complete questions. In contrast, elliptical assertions differ from elliptical questions in several aspects. They always contain the word ‘jak ’ in
initial position, they do not have question interpretation and cannot be reconstructed into full
clauses. I propose that elliptical multiple questions result from TP-deletion after movement of the
wh-words into CP domain (so-called multiple sluicing) and the TP-ellipsis is driven by the Focus
feature. This explains that it can also apply to referential focused DPs. As for elliptical assertions,
I claim that they are base-generated and involve semantic ellipsis, i.e. a null category, which gets
its interpretation from a TP in the previous context.
Keywords
Ellipsis; Wh-words; Multiple Questions; Sluicing; Czech.
1. Introduction
This paper deals with elliptical sequences of multiple wh-words, as in (1) and (2).
These sequences are elliptical in that they have sentential interpretation despite
the absence of a syntactically full clause. I argue however that they cannot receive
a uniform analysis. In particular, only sequences in (1) can be derived from full
57
ČLÁNKY – ARTICLES
64 / 2016 / 1
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
multiple questions by TP-deletion (Merchant 2001). In contrast, sequences in (2)
are base-generated and contain a null category e that receives its clausal content
by a semantic reconstruction (Chung et al. 1995).
(1)
Přenáší informace, ale nevím od koho komu.
(he) brings information, but (I) don’t know from whogen whodat (he brings information)
‘He brings information but I don’t know to whom he brings information, and
from whom.’
(2)
Prý jsou hotely u pláže už plné? Jak kde.
they say hotels on the beach are full? how where
‘It seems that the hotels on the beach are full. Some are and some are not.’
The paper is organized as follows. In Section 2, I present the properties of ellipsis
in questions with a single wh-word, known as sluicing. In Section 3, I compare the
properties of wh-words in multiple questions and in elliptical sequences. I show
that sequences in (1) share most properties with non-elliptical questions, while
those in (2) behave differently, in particular with respect to ordering and interpretation. Section 4 deals with the type of ellipsis in these sequences: deletion of
a syntactic structure driven by the focus feature in (1), and semantic reconstruction in (2). Section 5 sums up the paper.
2. Sluicing
Sluicing refers to ellipsis of an interrogative clause in which everything except for
the interrogative word was elided (Ross 1969). Sluicing appears both in embedded
contexts and in independent clauses related to a previous utterance, as shown in
(3). The antecedent of the elided clause can be found in independent utterances as
in (3b), and in coordinate and subordinate clauses, see (3a) and (4a) respectively.
It usually precedes the wh-remnant, but it can follow it in an appropriate context,
see (4b).
(3)
(4)
a. John bought something, but I don’t know what.
(= what John bought)
b. John bought something. What?
(= what did John buy?)
a.
Pokud chceš někoho pozvat, dobře si rozmysli koho.
if (you) want (to) invite someone, bethink well whoacc
b. (Ukázal jsem jí mapu): Nevím ještě kam, ale někam si určitě vyjedeme.
(i showed her the map): I don’t know where, but we will go out somewhere.
58
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
(5)
Domluvil si (s někým) práci v Tel Avivu, ale nevzpomínám si s kým. (ČNK)1
(he) arranged (with someone) a job in Tel Aviv, but (I) do not know with whom
(6)
a. Jan koupil někomu dárek, ale nevím komu / *koho. (komu / *koho koupil dárek).
John bought someonedat a gift, but (I) don’t know whodat / whoacc
b. Jan chce někoho pozvat, ale neví ještě koho / *komu. (koho / *komu chce pozvat).
(7)
a.
John wants (to) invite someoneacc, but (he) doesn’t know whoacc / whodat
Musel dostat od někoho informaci. A já se ptám, *(od) koho.
(he) must have received information from someone and I ask (from) whom
b. He was talking to somebody, and I want to know (to) whom.
The case-marking and the possibility (or not) of preposition stranding constitute two principal pieces of evidence for deriving sluicing by TP-deletion from
non-elliptical questions (Ross 1969, Lasnik 2001, Merchant 2001), as illustrated
in (8):
(8)
a. John bought something but I don’t know [CP what i [TP John bought t i ]].
b. Jan něco koupil, ale nevím [CP coi [TP Jan koupil t i ]].
Assuming the analysis in (8), I make hypothesis that the elliptical sequences we are dealing with are derived from questions with more than one
wh-word, so-called multiple questions (cf. Stjepanović (2003) for Serbo-Croatian,
Grebenyova (2006) for Russian). This hypothesis predicts (i) that the wh-remnants have the same properties as the wh-words in non-elliptical multiple questions, and (ii) that they can be syntactically reconstructed into full questions. In
the following section, I show that these predictions are only born out for wh-sequences like in (1).
1
ČNK is a shorthand for Czech National Corpus. In this paper, I use the subcorpus SYN2010
available on http://www.korpus.cz.
2
Contrary to cleft-sentences where the wh-word is in nominative case (cf. Grebenyova 2006).
59
ČLÁNKY – ARTICLES
In languages with morphological case-marking, the wh-remnant must bear the
same case as its antecedent or as the wh-word in non-elliptical questions2, see (6).
The wh-remnant must also be accompanied by the same preposition as its correlate, in contrast to languages like English that allow preposition-stranding, see (7).
64 / 2016 / 1 The sentence containing the antecedent of sluicing also contains an explicit or an
implicite correlate of the wh-word, corresponding to an indefinite DP or AdvP, like
někdo, něco, někde, etc.:
ČLÁNKY – ARTICLES
64 / 2016 / 1
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
3. Properties of wh-words in multiple questions and
in elliptical sequences
There exists substantial literature about the properties of multiple wh-words in
Slavic.3 In this section, I compare wh-words in multiple questions and in elliptical
sequences, focusing on their adjacency, ordering, interpretation and combinability. I show that elliptical wh-sequences fall into two groups: sequences that behave
like multiple questions4 , and sequences with an initial jak (‘how’) that behave differently, in particular with respect to their ordering and interpretation.
3.1 Adjacency
All wh-words in informative multiple questions in Czech must be fronted. They can
however be separated by second position clitics (auxiliary verbs and pronouns), as
in (9). The clitics’ position is related to interpretation, see Section 3.3.
(9)
a. Co komu jsi koupil? / Co jsi komu koupil?
what whodat cl.2sg you bought / what cl.2sg whodat bought
‘What did you buy, and for whom?’
b. Rád bych věděl, kdo se kde schovává / ?kdo kde se schovává.
I wonder who cl.refl where hides / who where cl.refl hides.
‘I wonder who is hiding, and where.’
In contrast, wh-remnants in elliptical sequences must always be adjacent, see (10).
Assuming that clitics constitute a barrier between TP and CP domains, the ellipsis
site in wh-sequences must involve TP and clitics.
(10) a. Prý jsi každému něco koupil. Řekni mi co komu / *co jsi komu.
they say (you) cl.2sg bought something to everyone tell me what whoDAT /
what cl.2sg whoDAT
‘It seems that you bought someone something. Tell me what you bought, and
to whom.’
b. Každý se někde schovává, ale nevím kdo kde / *kdo se kde.
everyone cl.refl somewhere hides, but I don’t know who where / who
cl.refl where
‘Everyone is hiding somewhere, but I don’t know who, and where.’
3E.g. Rudin (1988), Bošković (1999, 2002), Meyer (2002, 2004), Gruet-Skrabalova (2011).
4
With some differences that will be explained in Section 4.
60
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
3.2 Ordering and correlates
(11)
a. Co komu koupil? / Komu co koupil?
what whodat (he) bought / whodat (he) what bought
64 / 2016 / 1 We observe no superiority effects in multiple questions in Czech, whether they
are embedded or not (Meyer 2004):5
b. Řekni mi, kdo koho / koho kdo pozval k tanci.
Although both orders are possible in elliptical sequences, parallel ordering of whremnants with respect to their correlates is often preferred, see (12b).6 The correlates of multiple wh-remnants correspond to quantified and indefinite XPs as in
(12), or to plural referential DPs, as in (13), and the wh-remnants always bear the
case of their correlates.
(12) a. Každému něco koupil, ale komu co / co komu, to nevím.
everyoneDAT something (he) bought, but whodat what / what whodat, I don’t
know
b. Každý pozval někoho k tanci, ale nevzpomínám si kdo koho / ??koho kdo.
c.
(13) a.
everyone invited someone to dance but I don’t remember who whoacc /
whoacc who
Spojka [...] přenáší informace, ale nevím od koho komu. (ČNK)
(the liaison officer) brings information, but (I) don’t know from whogen
whodat
Domluvili se, že je budou střílet do srdce. A rozdělili si, kdo koho. (ČNK)
(they) arranged to shoot them into the heart. And they decided who whoacc
b. Moji kamarádi se na tu oslavu (nějak) přestrojili, ale nevzpomínám si už, jak kdo
my friends for the party (somehow) dressed up, but I don’t remember how
who
On the contrary, sequences introduced by jak (‘how’) as in (14) behave differently
because (i) the order of the wh-remnants is strict, and (ii) the wh-remnants do not
have indefinite nor quantified correlates. So, contrary to (13b) where jak has an
indefinite correlate nějak, jak in (14) is linked to specific adverbials (of different
semantic types) in the antecedent clause, i.e. full, nicely, main, very. The wh-word
5
6
In the corpus (ČNK, subcorpus SYN2010), jak appears much more frequently in the initial position.
According to Grebenyova (2006), the parallelism is obligatory in Russian.
61
ČLÁNKY – ARTICLES
tell me who whoacc / whoacc who invited to dance
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
ČLÁNKY – ARTICLES
64 / 2016 / 1
following jak always has a plural referential correlate, generally implicit, which
bears the same case, see (14b,c).
(14) a. Prý jsou hotely u pláže už plné? Jak kde / *kde jak. (ČNK)
they say hotels on the beach are full? how where / where how
b. Je pravda, že se tady (lidem) příjemně tráví čas? Jak komu / *komu jak.
c.
is (it) true that (people DAT) have a nice time here? how whodat / whodat how
Bylo vaší hlavní inspirací Chile? Záleží jak v čem / *v čem jak.
was your main inspiration Chile? it depends how in what / in what how
3.3 Distribution
Multiple questions can be both independent and embedded, see (11) above. In
comparison, elliptical sequences mostly appear embedded under predicates that
subcategorize an interrogative clause, except for jak-sequences in (14). These latter
mostly constitute independent utterances, or appear embedded under the predicate záleží (‘it depends on’), which is due to their interpretation (see Section 3.4).
Furthermore, elliptical sequences usually follow their antecedent clause.
However, we can find sequences nevím jak kdo (‘I don’ t how who’) preceding their
antecedent, as in (15a), in which the wh-word kdo contrasts with an explicit referential correlate and can be itself replaced by another referential DP. When the
antecedent precede the wh-remnants, kdo refers to each individual included in the
correlate, thus I don’ t know how each one of us was satisfied in (15b). This order is
therefore impossible when the correlate denotes a single individual, compare (16a)
and (16b). In Section 4, I show that these sequences can receive the same syntactic
analysis as other sequences embedded under ‘to know’.
(15)
a. Nevím, jak kdo / jak vaši kolegové, ale my jsme byli velmi spokojeni. (ČNK)
(I) do not know how who / how your colleages but we were very pleased
b. My jsme byli velmi spokojeni, ale nevím, jak kdo.
(16) a.
Nevím, jak koho / jak Pavla, ale mě docela rozčiluje, když... (ČNK)
(I) do not know how whoacc / how Paulacc but (it) quite irritates me when...
b. *Mě docela rozčiluje, ale nevím jak koho, když....
3.4 Interpretation
Multiple questions can have pair-list (PL) reading and specific-pair (SP) reading. In Czech, non-adjacent wh-words require pair-list (distributive) reading and
conjoined wh-words require specific reading, see (15). In absence of clitics, the
62
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
(17)
a. Kdo se komu omluvil? Jan se omluvil Petrovi, Pavel Tomášovi a Martin Ondrovi. (PL)
who cl.refl whoDAT apologized? J. cl.refl apologized to P., P. to T. and M. to O.
64 / 2016 / 1 question is compatible with both readings (Gruet-Skrabalova 2011). Adjacent
and conjoined wh-items can also have contrastive reading if the participants are
known and the question only asks to identify their respective roles.
b. Kdo komu se omluvil? Jan se omluvil Petrovi (a ne Petr Janovi). (SP)
c. Kdo a komu se omluvil? Jan se omluvil Petrovi (a ne Petr Janovi). (SP)
who whoDAT cl.refl apologized? J. cl.refl apologized to P. (and not P. to J.)
Elliptical sequences in (12) and (13) can be reconstructed into full multiple questions, as shown in (18), and are interpreted like that. Their specific interpretation
depends on the correlates of wh-remnantes, see (19): quantified correlates require distributive reading and therefore adjacent wh-remnants, while indefinite
correlates require specific reading and occur more frequently with conjoined7
wh-remnants.
(18) a. Každému něco koupil, ale komu co koupil, to nevím.
b. Každý pozval někoho k tanci, ale nevzpomínám si kdo koho pozval k tanci.
c.
(19) a.
Moji kamarádi se na tu oslavu přestrojili, ale nevzpomínám si už, jak kdo se na tu
oslavu přestrojil.
Každý se někde schovává, ale nevím kdo kde. / *kdo a kde. (PL)
everyone cl.refl somewhere hides, but I do not know who where / who and
where
b. Někdo se někde schovává, ale nevím ??kdo kde / kdo a kde. (SP)
someone cl.refl somewhere hides, but I do not know who where / who and
where
c.
Věděl, že někdo někomu napsal, ale kdo komu / kdo a komu, to nevěděl. (SP)
(he) knew that someone someoneDAT wrote, but who (and) whoDAT (he) didn’t
know
In contrast, elliptical jak-sequences in (14) are interpreted as assertions. They imply that a previous assertion, given or presupposed in the context, is true for some
x (individuals, objects, moments, etc.), but not for all, as shown by the paraphrase
and the context in (20). Thus, they always have distributive reading. The type of
7Cf. Tak se mi zdá, že se tady někdo někomu líbí. No to by mě teda vážně zajímalo, kdo a komu. (ČNK)
it seems to me that here someone someoneDAT pleases I really wonder who and whodat
63
ČLÁNKY – ARTICLES
who whoDAT cl.refl apologized? J. cl.refl apologized to P. (and not P. to J.)
64 / 2016 / 1
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
x is given by the wh-word following jak: x is an individual from the set kdo (‘who’)
in (20a), and a moment from the set kdy (‘when’) in (20b). Jak behaves here therefore not as an interrogative word, but as an adverbial quantifier that distributes
the affirmative or the negative value of the preceding assertion over the set of
x given by the subsequent wh-word. The syntactic reconstruction in (21) is impossible, because the deleted TP should be both affirmative and negative.
(20) a. Je pravda, že se tady příjemně tráví čas? Jak komu. = Někomu ano, někomu ne.
ČLÁNKY – ARTICLES
is (it) true that (one) has a nice time here? how whodat = some have, and
some have not
b. Za co utrácíš kapesné? Jak kdy, někdy to utratím, někdy šetřím.
for what (you) spend money? how when, sometimes (I) spend it, sometimes
(I) spare it
(21) a.
Je pravda, že se tady příjemně tráví čas? *Jak komu se tady (příjemně) tráví čas.
b. Za co utrácíš kapesné? *Jak kdy (za co) utrácím kapesné.
3.5 Combinability
Although all wh-words can appear in multiple questions, two adjacent adjuncts
are usually considered as less felicitous in comparison to conjoined adjuncts.
This is however due to the possibility or not to obtain a distributive reading (see
above): in (22), distributive reading of kdy kde (‘when where’) is possible only in the
example (b): ‘for each period of holidays, where will you be?’. In (23), distributive
reading of jak kdy (‘how when’) also better obtains in (b): ‘he does not know for
each party how his girlfriends were dressed’.
(22)
a. *Kdy kde / Kdy a kde se máte setkat?
when where / when and where should you meet
b. Řekni mi kdy kde / kdy a kde budete během prázdnin.
tell me when where / when and where (you) will be during holidays
(23)
a.
??Jak kdy / Jak a kdy hodnotíte studenty na univerzitě?
how when / how and when (you) evaluate students at the university
b. Neví vůbec, jak kdy byly oblečeny jeho přítelkyně. (ČNK)
(he) does not know how when were dressed his girlfriends
In my corpus study of 16 combinations8, I have found almost 1.800 occurrences of
multiple questions, the five most frequent combinations being co kdo (253), co kdy
8
My corpus study focused on 16 combinations of the following wh-words in both orders and all
cases: kdo (who), co (what), kdy (when), kde (whereloc), kam (wheredir), jak (how).
64
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
3.5 Summary
The properties discussed in this section show that elliptical wh-sequences fall
into two groups: those that share the properties of multiple questions (that I call
E-sequences) and those that do not ( jak-sequences):
Tab. 1. Properties of multiple questions and elliptical sequences
Adjacency
Interpretation
Free
order
Contexts
Correlates
Combinations
Reconstruction
Multiple
questions
not necessarily
question
distributive /
specific
yes
main or
embedded
–
all
–
E-sequences
ex. (1)
yes
question
distributive /
specific
+ contrastive
yes
mostly embedded
under ‘to know’
indefinites,
quantifiers,
referential
XPs
all
but less acceptable
yes
jak-sequences
ex. (2)
yes
assertion
distributive
no
mostly main or
embedded under
‘to depend’
specific
adverbial +
referential
XPs
jak
+ wh-word
no
4. Two types of ellipsis
In this section, I propose that elliptical sequences involve two different types
of ellipsis: E-sequences involve syntactic deletion in multiple questions, while
jak-sequences contain a null base-generated category e.
4.1 Deletion in multiple questions
Wh-remnants in E-sequences behave as wh-words in multiple questions. Consequently, E-sequences can receive the syntactic analysis of sluicing constructions
9
Sequences of conjoined adjuncts are also very frequent.
10
I do not consider the lexicalized sequence kdo s koho (‘who of whogen’) meaning ‘who of them two
will prevail over’: Den konečného zúčtování se přiblížil. Nyní se ukáže, kdo s koho. (ČNK) ‘The day of final
judgement came close. Now it will reveal who will have the upper hand.’
65
ČLÁNKY – ARTICLES
(24) Líp ti to neřeknu a nevím, zda kdy kdo.
(I) cannot say it better to you and I do not know whether when who.
‘I cannot say it better, and I don’t know whether someone could at some other
moment.’
64 / 2016 / 1 (223), jak co (188), kdo kdo (163), a kdo co (137). I have not found any occurrence of
sequences kdy jak, kde kam / kam kde, kde jak, and kam jak / jak kam. In contrast, we
have found less than 100 elliptical sequences9, the most frequent being independent jak-sequences (88): jak kdo (67), jak co and jak kdy / jak kde. I have only found 510
elliptical sequences with other wh-words than jak. In one of them, the wh-words
have indefinite interpretation (see Section 4.1):
ČLÁNKY – ARTICLES
64 / 2016 / 1
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
discussed in the Section 2: they contain a clausal structure, the wh-words move to
the left-periphery, and TP is deleted under identity with the TP in the antecedent
clause (Merchant 2001):11
(25) Každý někoho pozval, ale nevím [CP kdo1 koho2 [TP t1 pozval t 2]].
In Merchant’s (2001) analysis, TP-deletion is licensed by the head C which bears
the features [+Q] and [+Wh]. It has been however argued (Bošković 1998, 2002,
Stepanov 1998, Stjepanović 2003) that movement of multiple wh-words in Slavic
is driven by the Focus feature, and that multiple sluicing is licensed by every head
bearing a Focus feature (Grebenyova 2006). I propose that a focus-driven analysis of sluicing is appropriate for multiple sluicing in Czech, because it allows to
explain adjacency and specific, or at least contrastive, interpretation of wh-remnants (see Section 3). I assume that only adjacent wh-words in multiple questions
appear both in the CP domain, i.e. before clitics, and occupy the specifiers of IntP
and FocP respectively (Rizzi 1997, Lenertová 2001, Gruet-Skrabalova 2011), see
(26). Consequently, sluicing licensed by the head Focus will results in sequences
of adjacent wh-remnants with specific or contrastive interpretation. The elided
structure corresponds to TP and clitics, thus FinP. Semantically, the contrast obtains more easily with the wh-words of the same type, which explains the preference for elliptical sequences kdo kdo.12
(26)
a. [ForceP [IntP Kdo [FinP se [TP komu [vP t omluvil t]]]]]?
=> *..., ale nevím [ForceP [IntP kdo [FinP se [TP komu [vP t omluvil t]]]]]
b. [ForceP [IntP Kdo [FocP komu [foc] [FinP se [TP [vP t omluvil t ]]]]]]?
=> ale nevím [ForceP [IntP Kdo [FocP komu [foc] [Fin se [TP [vP t omluvil t ]]]]]]
Finally, focus-driven analysis of sluicing predicts that sluicing should also be possible after focused referential DPs (Grebenyova 2006). This is actually true for
both a single DP and a sequence of two DPs, as shown in (27). Thus, the proposed
analysis can also account for sequences combining wh-words, referential DPs and
indefinite pronouns, as long as they are contrastively focused, see (28).
(27)
a. Mohl to někdo udělat sekerou? Spíš bych řekl, že mačetou. (ČNK)
could someone do it (with) axeinstr? I would say that (with) truncheoninstr
11
The identity of ellipsis does not concern agreement features (Merchant 2001, 2006).
12
This also explains that three and more wh-remnants are better conjoined or separated by
a comma.
66
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
b. Že Češi spolupracují s Francouzi je jisté, ale jestli Francouzi s Čechy už tak jisté
není.
(28)
a. Každý s někým tancoval, ale kdo s Evou (tancoval), na to si už nevzpomínám
64 / 2016 / 1 that Czechs cooperate with French is sure, but that French with Czechs is
not so sure
everyone with someone danced but who with Eva (danced) I don’t remember
(I) don’t know how whoACC / how Paul ACC (it irritates) but (it) certainly irritates me
c.
Líp ti to neřeknu a nevím, zda kdy kdo (ti to řekne líp).13
(I) cannot say it better and I don’t know whether someone could say it better
at another moment
4.2 Semantic reconstruction
Jak-sequences do not allow syntactic reconstruction, although they have clausal
interpretation. These non-sentential assertions confirm the value of a previous
(presupposed) assertion for some but not all x in a set given by the wh-word following jak. The variable x has an implicit correlate in the presupposed assertion, see
(29a). The syntactic reconstruction is impossible because these sequences imply
both an affirmative and a negative clause, see (29b) and Section 3.4.
(29)
a. Je pravda, že se tady (lidem) příjemně tráví čas? Jak komu.
is (it) true that (peopleDAT) have a nice time here? How whodat
b. Někomu se tady tráví čas příjemně, někomu ne.
some have a nice time here, some have not
I suggest that jak-sequences are base-generated and contain a null category [e] that
is semantically reconstructed at LF as other anaphors (Chung et al. 1995, Lobeck
1995). The fact that the interrogative DP following jak must bear the same case as
its correlate is necessary to ensure its syntactic and semantic integration into the
ellipsis site (Chung 2013, 29). More detailed semantic analysis of these constructions is however beyond the scope of this paper.
(30)
a. Je pravda, že se tady příjemně tráví čas? [CP Jak komu [TP e]].
b. LF: [CP Jak komui [TP lidemi se tady příjemně tráví čas]]
13
We assume that the polarity feature of the clause is checked in the CP (LAKA 1990), thus not relevant for TP identity.
67
ČLÁNKY – ARTICLES
b. Nevím, jak koho / jak Pavla (to rozčiluje), ale mě to teda rozčiluje.
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
ČLÁNKY – ARTICLES
64 / 2016 / 1
5. Conclusion
In this paper, I dealt with elliptical sequences of multiple wh-words in Czec. I argued that they involve two different constructions, elliptical multiple questions
and elliptical assertions. I first showed that wh-words in elliptical multiple questions share the properties of wh-words in non-elliptical questions: they have the
same syntactic distribution and semantic interpretation, they do not show superiority effects, and their combinability depends on their reading. Moreover, elliptical questions can be reconstructed into syntactically complete questions. In contrast, elliptical assertions differ from elliptical questions in several aspects. They
always contain the word jak in initial position (the order of wh-words is therefore
not free), they don’ t have question interpretation and cannot be reconstructed
into full clauses. I then proposed that elliptical multiple questions result from TPdeletion after movement of the wh-words into CP domain (so-called multiple sluicing) and that the TP-ellipsis is driven by the focus feature. Thus explains that it
can also apply to referential focused DPs. As for elliptical assertions, I claimed that
they are base-generated and involve semantic ellipsis, i.e. a null category, which is
interpreted anaphorically with respect to a TP in the previous context.
REFERENCES
Bošković, Željko. 1999. On multiple feature-checking: Multiple wh-fronting and multiple
headmovement. In: Epstein, Samuel – Hornstein, Norbert, eds. Working Minimalism.
Cambridge, Mass.: MIT Press, pp. 159–187.
Bosković, Željko. 2002. On multiple wh-fronting. Linguistic Inquiry. 33, pp. 351–384.
Chung, Sandra – Ladusaw, William – McCloskey, James. 1995. Sluicing and logical form.
Natural Language Semantics. 3, pp. 239–282.
Chung, Sandra. 2013. Syntactic Identity in Sluicing: How Much and Why. Linguistic
Inquiry. 44(1), pp. 1–44.
Grebenyova, Lydia. 2006. Sluicing and Multiple Wh-fronting. In: Richa, Srihti – Sinha,
Samar, eds. Proceedings of GLOW in Asia. Vol. 5. pp. 219–242. Also available at: http://
engagedscholarship.csuohio.edu/cgi/viewcontent.cgi?article=1001&context=ling_
facpub.
Gruet-Skrabalova, Hana. 2011. Czech questions with two wh-words. In: Kosta, Peter –
Schürcks, Lilia, eds. Formalization of Grammar in Slavic Languages. Frankfurt am Main:
Peter Lang, pp. 179–192.
Laka, Itzar. 1990. On the syntax of negation. Cambridge, Mass. Ph.D. thesis. MIT.
Lasnik, Howard. 2001. When you can save a structure by destroying it? In: Kim, Minjoo –
Strauss, Uri, eds. Proceedings of NELS 31. Amherst, Mass.: GLSA, pp. 301–320.
68
Hana Gruet-Skrabalova
Ellipsis in Sequences of Multiple Wh-Words
UBP & LRL (EA 999), Clermont University
29 boulevard Gergovia, 630 00 Clermont-Ferrand
France
[email protected]
69
ČLÁNKY – ARTICLES
Hana Gruet-Skrabalova
64 / 2016 / 1 Lenertová, Denisa. 2001. On Clitic Placement, Topicalization and CP-Structure in Czech.
In: Kosta, Peter – Frasek, Jens, eds. Current Approaches to Formal Slavic Linguistics.
Frankfurt am Main: Peter Lang, pp. 294–305.
Lobeck, Anne. 1995. Ellipsis: functional heads, identification, and licensing. Oxford: Oxford
University Press.
Merchant, Jason. 2001. The Syntax of Silence: Sluicing, islands, and the theory of ellipsis.
Oxford: Oxford University Press.
Merchant, Jason. 2004. Fragments and Ellipsis. Linguistics and Philosophy. 27(6), pp. 661–
738.
Merchant, Jason. 2006. Sluicing. In: Everaert, Martin – van Riemsdijk, Henk, eds. The
Blackwell Companion to Syntax. Oxford: Wiley-Blackwell, pp. 269–289.
Meyer, Roland. 2002. On multiple wh-fronting and wh-clustering in Czech. In: Brown,
Wales et al., eds. Formal Approaches to Slavic Linguistics 11: The Amherst Meeting. Ann
Arbor: Michigan Slavic Publications, pp. 393–412.
Meyer, Roland. 2004. Superiority effects in Russian, Polish, and Czech. Cahiers linguistiques d’Ottawa. 32, pp. 44–65.
Rizzi, Luigi. 1997. The Fine Structure of the Left Periphery. In: Haegeman, Liliana, ed. Elements of grammar. Dordrecht: Kluwer, pp. 281–337.
Ross, John. 1969. Guess who? In: Binnick, Robert et al., eds. Papers from the 5th regional
Meeting of CLS. Chicago: Chicago Linguistic Society, pp. 252–286.
Rudin, Catherine. 1988. On multiple questions and multiple Wh-fronting. Natural Language and Linguistic Theory. 6(4), pp. 445–501.
Stepanov, Arthur. 1998. On Wh-Fronting in Russian. In: Tamanji, Pius – Kusumoto,
Kiyomi, eds. Proceedings of NELS 28. Amherst, Mass.: GLSA, pp. 453–467.
Stjepanović, Sandra. 2003. Multiple Wh-Fronting in Serbo-Croatian Matrix Questions
and the Matrix Sluicing Constructions. In: Boeckx, Cedric – Grohmann, Kleanthes,
eds. Multiple Wh-fronting. Amsterdam: John Benjamins, pp. 255–284.